Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Stressforskningsinstitutet
Stressforskningsrapport nr 308
Work hour flexibility and the ability to sustain working life to retirementTorbjörn Åkerstedt, Michael Ingre & Claire Eriksen
Nr 308 2003
Work hour flexibility and the abilityto sustain working life to retirement
Torbjörn ÅkerstedtMichael IngreClaire Eriksen
Work hour flexibility and the ability to sustain working life to retirement
Torbjörn ÅkerstedtMichael IngreClaire Eriksen
Institute for Psychosocial Medicine & Karolinska Institutet
Stockholm, Sweden
Editor-in-Chief: Töres Theorell
Psychosocial Medicine
is the comprehensive term for interdisciplinary research on various human en-vironments,how they are perceived and experienced and their influence on the human organism bothpositively and negatively.
Our environment abounds with psychosocial risk situations. Many of them may cause
emotional as well as physical disorders, dysfunctions and/or diseases. The goal of psycho-social medical research is to study the influence of these risk situations, on the one hand, and
man´s emotional, behavioural and physiological reactions on the other, as well as their
relation to mental and somatic morbidity and mortality. Our research is therefore
interdisciplinary and comprises experimental studies both in the laboratory and in real life, as
well as epidemiological surveys and analyses.
In Stockholm this research is conducted through the unique collaboration of:
• IPM - The National Swedish Institute for Psychosocial Factors and Health,• the Division of Psychosocial Factors and Health at the Karolinska Institutet,• WHO´s Psychosocial Centre and
• the Centre for Suicide Research and Prevention at the Stockholm County Council.
IPM and the Division of Psychosocial Factors and Health, Karolinska InstitutetDirector: Töres Theorell, Professor
Units / research programs at IPM:Child and Trauma
Head: Frank Lindblad, Associate Professor
Migration and Health
Head: Solvig Ekblad, Associate Professor
Psychosocial Factors and Health
Head: Töres Theorell, Professor
Stress – Prevention and Treatment
Head: Aleksander Perski, Associate Professor
Suicide Research and Prevention
Head: Danuta Wasserman, Professor
Work and Health / Sleep and Fatigue
Head: Torbjörn Åkerstedt, Professor
Copyright © 2003 Publisher and authors
ISSN 0280-2783
Akademitryck, Edsbruk; 2003
3
IntroductionWork hours have been shown to influence health, particularly night work, but
also long work hours (Costa, 1998; Åkerstedt, 1998; Knutsson and Bøggild,
2000). Another aspect of work hours is flexibility. Flexibility has during the
second half of the last century mainly referred to “flextime”, that is the ability to
vary the time for starting or finishing work according to the individual needs
(Wade, 1974). Well-being seems improved but the amount of systematic studies
is low (Torsvall and Åkerstedt, 1977; Tepas, 1985; Kogi, 1995; Kogi, 1997).
Also in connection with shift work the effects seem positive (McEwan Young,
1980; Knauth et. al., 1984). Sickness absence has been found to decrease in at
least two studies (Golembiewski et. al., 1974; Narayanan and Nath, 1982). On
the whole, however, there is very little systematic research on the connection
between influence on work hours and health.
Although not based on individual choice, there is also the employer’s flexibility
in using overtime, adding a second or a third shift, or using annual work hours.
At least overtime and night work has been the focus of much research and will
not receive any emphasis here. They will, however, be included in the analyses.
Initially, it was observed that there is a strong tendency towards a type of
flexibility of work hours that requires the worker to extend or reduce the amount
of work hours per week depending on the needs of production. At present we
know too little of the effects of such arrangements but the there does not seem to
be any advantage for the individual in this type of arrangement, rather the
opposite since it reduces the influence of the individual (Knauth, 1998).
However, such disadvantages may be compensated for by increasing the
freedom of the individual to choose when to work
The present paper is part of a larger project on flexible work hours within the
framework of the SALTSA program (Costa et. al., in press). It uses as dependent
variable the self-reported ability to remain in the work force until the normal
time of retirement, short and long-term sick leave. It relates this variables to the
perceived influence on work hours but also on other factors related to employer
flexibility such as shift work, overtime work and others. A particular effort was
made to control for as many background and work-related variables as possible.
4
TABLE OF CONTENTS
INTRODUCTION.....................................................................................................3
SUMMARY ..............................................................................................................5
SAMMANFATTNING ............................................................................................6
METHOD ..................................................................................................................7
SAMPLE AND POPULATION ......................................................................................7
DROP-OUTS .............................................................................................................7
QUESTIONNAIRE......................................................................................................8
Dependent variables ..........................................................................................8Background factors............................................................................................9Work-related psychosocial factors (the demand-control-support model) ......9Work hour factors ............................................................................................10
ANALYSIS..............................................................................................................10
RESULTS................................................................................................................11
QUIT WORKING BEFORE THE NORMAL AGE OF RETIREMENT .................................11
QUIT WORKING DUE TO MENTAL STRAIN ..............................................................13
Short-term sick leave .......................................................................................15Long-term sick leave........................................................................................17
DISCUSSION .........................................................................................................19
REFERENCES........................................................................................................22
5
SummaryÅkerstedt, T., Ingre, M., Eriksen, C. Work hour flexibility and the ability to
sustain working life to retirement. Stress Research Report no 308, Stockholm
2003. ISSN 0280-2783
Irregular work hours, particularly night work, have negative effects on health.
Another work hour aspect, flexibility, has seen very little research but should
logically have positive effects on health. The present paper is part of a larger
project on flexible work hours within the framework of the SALTSA program. It
uses the self-reported ability to remain in working life until normal time of
retirement as dependent variable and investigates its relation to influence on
work hours, other flexibility-related factors such as shift work, overtime and
others, while controlling for as many background and work-related variables as
possible. The main investigation involves randomly selected men and women,
who on October 1 2000 were between that ages of 25-75 years old and were
registered as Swedish citizens. The total number of respondents was 3493 (53%
response rate).
For this purpose was used logistic regression analysis. In the first step a series of
crude regression analyses were carried out using background variables,
psychosocial work indicators and work hour variables. The dependent variables
were the self-estimated ability to remain in the working force until normal
retirement; the self-estimated inability to remain in the working force because of
too high demands, number of sickdays exceeding 24 the last year, and being on
a sick leave that had lasted in excess of 3 months. In a second step the analysis
was controlled for background factors and in a third step also for psychosocial
work factors.
The main impression of the results is that work hours seem to be important for
all the variables analyzed in this study, but to a varied degree. Part-time work
showed increased risk for short-term sick leave and quitting work before normal
retirement. Temporary employment status was associated with an over risk for
long-term sick leave but, an under risk for short-term sick leave. Irregular work
hours with night work and work hours with varied shift length were associated
with quitting work due to mental strain. Low influence over work-hours was
associated with a higher risk for long-term & short-term sick leave and to quit
work due to mental strain. Low influence over work hours also indicated an
under risk for quitting work before normal age of retirement. Overtime work
seems to be mainly associated with an under risk for sick leave. The
combination of overtime without interest or with force increased, however, the
risk of quitting work before normal age of retirement. Unpaid overtime
increased the risk for quitting work before normal retirement but paid overtime
decreased the risk.
6
SammanfattningÅkerstedt, T., Ingre, M., Eriksen, C. Flexibla arbetstider och förmågan att stanna
i arbetsliver fram till pensionering. Stressforskningsrapport nr 308. Stockholm
2003. ISSN 0280-2783
Oregelbundna arbetstider, speciellt nattarbete, påverkar hälsan negativt. En
annan arbetstidsaspekt, flexibilitet, har inte forskats kring i någon större
utsträckning men borde rimligen ha positiva effekter på hälsan. Föreliggande
rapport är en del av ett större projekt under SALTSAS (SACO, TCO, LO) ram. I
denna studie används självrapporterad förmåga att kvarstå i arbetslivet till
pensionen som beroende variabel (även sjukfrånvaro analyseras). Dessa
variabler ställs i relation till inflytande över arbetstiden, men även andra faktorer
som övertidsarbete eller skiftarbete. Dessutom kontrolleras för en rad bak-
grundsvariabler och psykosociala arbetsmiljövariabler. Undersökningen
omfattar 3493 personer mellan 25 och 75 års ålder, utgörande ett slumpmässigt
urval av svenska medborgare. Antalet svarande motsvarar en svarsfrekvens på
53%.
Data analyserades med hjälp av logistisk regressionsanalys i flera steg. I det
första steget gjordes en analys utan hänsyn till andra oberoende variabler
(“crude”). I ett andra steg kontrollerades för bakgrundsvariabler och i ett tredje
steg kontrollerades också för psykosociala arbetsmiljövariabler (krav,
inflytande, socialt stöd m.fl.).
Resultaten visar att arbetstiderna hade betydelse för den självrapporterade
förmågan att kvarstå i arbetslivet fram till pensionsåldern samt för
sjukfrånvaron. Deltidsarbete visade en ökad risk för sjukfrånvaro och att sluta
arbeta före pensionsåldern. Att vara temporärt anställd visade på en överrisk för
långtidssjukskrivning men en underrisk för korttidssjukfrånvaro. Oregelbundna
arbetstider visade en ökad risk att sluta före pensionsåldern på grund av mental
påfrestning. Lågt inflytande över arbetstiderna var associerat med en underrisk
att sluta arbeta i förtid. Övertid var i huvudsak associerat med en underrisk för
sjukfrånvaro men i kombination med tvång eller avsaknad av lust var övertid
associerat med en överrisk att sluta arbeta före pensionsåldern. Obetald övertid
ökade risken och betald övertid minskade risken att sluta arbeta i förtid.
7
Method
Sample and population
The sample was made up of randomly selected men and women, who on
October 1 2000 were between the ages of 25 and 75 years and were registered as
Swedish citizens. A selection from the Statistics Sweden (SCB) was drawn in 20
strata, defined by sex, age (5 groups), and place of residence (2 groups).
The age groups were 25-34 years, 35-44 years, 45-54 years, 55-64 years, and
65-75 years. The place of residence was divided into an urban and a rural area.
A person was considered to be a resident in an urban area if he was registered as
living on a property in a place with a connected set of buildings with at the most
200 meters apart and with at least 200 inhabitants. The remainder was
considered residents of rural areas.
Within each stratum there was a random selection of 340 individuals, that is, a
total of 6800 individuals. The number of individuals in each stratum was based
on an estimated participation frequency of 60 percent and a target of 200 people
per stratum.
SCB made the selection as well as the data collection. The latter was done by
questionnaire mailed to the participants during October 20 2000 – February 23
2001. Reminders were sent out on two occasions to those who had not filled out
the questionnaire. The questionnaire was enclosed with the second reminder.
Variables identifying individuals were removed after closing the collection
procedure.
Drop-outs
163 individuals that did not fit the criteria for age were removed. 6637 (3306
men and 3331 women) remained for analysis. The questionnaire was returned by
3493 (53 percent) individuals, 1631 (49 percent) of the men and 1862 (56
percent) of the women. In a previous report the drop-outs were analyzed. The
women answered to a greater extent than the men, older people (45 years and
older) more so than young people (under the age of 45), and residents of rural
areas to a greater extent than residents of urban areas. The proportion of people
who did not have an income (tax- year 1999) was lower among the ones who
filled out the questionnaire than the ones who did not. This difference was more
prominent among the women (25 percent filled it out, 38 percent did not) in
comparison with the men (24 percent filled it out, 28 did not). The size of the
taxable income for 1999 only differed marginally between those who filled out
the questionnaire and those who did not among the men (filled out: SEK
216798, did not fill out: SEK 213594), but it differed somewhat more among the
women (filled out: SEK 160066, did not fill out: SEK 143574).
8
The rate of participation differed also in relation to marital status and country of
birth. Regarding marital status the questionnaire was filled out to the highest
extent by married people (58 percent) and to the lowest extent by unmarried
people (46 percent) whereas divorced and widows/widowers fell in between.
Regarding place of birth the questionnaire was filled out by 54 percent that were
born in Sweden and by 44 percent (305 subjects) that were born abroad.
This report only includes people who belonged to the active workforce at the
time they filled out the questionnaire. This means that only those who stated that
they currently were employees or self-employed were included in the analyses.
A total of 1071 women and 1076 men (mean age 46 years) were analyzed.
A special case constituted the analyses of long-term sick leave (over 3 months).
A total of 100 participants were on a long-term sick leave however, the majority
(n=70) did not state that they were currently active in the workforce. The latter
were added to the analyses of long -term sick leave, which then included a total
of 2217 participants (1125 women and 1092 men, mean age 46 years).
Questionnaire
The questionnaire included a total of approximately 300 sub-questions.
Reference to a particular question in the questionnaire is given with the number
of the question within a parenthesis. Most of the questions concerned working
life and the participants were asked to base their answers on their current job, or
on their latest job if they were not working at the time they filled out the
questionnaire. In the following text, only those variables that were of current
interest for this report are described.
Dependent variables
1) Quit working before the normal age of retirement. The respondent was
asked to take a position on the statement “I will probably work until the normal
age of retirement”. Those who answered with a 4 or a 5 on a five-point scale
(1=“agree fully”, 5=“disagree completely”) were defined as those who do not
think that they will work until the age of 65 (n=479). Those who answered with
the alternatives 1, 2 or 3 constitute the control group (n=1550).
2) Quit working due to mental strain? These results are based on the
question: “My current work is too mentally/psychologically straining for me to
be able to continue until the normal age of retirement” (1=“agree fully”,
5=“disagree completely”). Those who answered with a 1 or a 2 constitute cases
(n=415). Those who had answered with a 3, 4 or a 5 constitute the reference
group (n=1587).
3) Short-term sick leave. Participants were asked to estimate their total
absence from work due to sickness during the last 12 months. Response
alternatives were “no day”, “1-7 days”, “8-24 days”, “25-99 days” and “100-365
days”. Those who had indicated 25 days or more constitute cases (n=175 and
those who indicated up to 24 days constitute the reference group (n=1843).
9
4) Long-term sick leave. Participants that had indicated that they currently
were on sick leave for more than three months constitute cases (n=100) and the
rest constitute the reference group (n=2117).
Background factors
Variables describing age, gender, marital status and having children under16 years of age were used in the analyses to control for background factors
together with variables describing different aspects of socio-economic status (se
below).
Education. One question asked for the respondent’s highest education.
Three groups were formed: elementary school, high school and university level.
Sector. One question regarded different sectors on the labor market:
manufacturing, health care, service, administration and research/development.
Employer . This variable describes if the respondent was self-
employed/employed in the private sector, by the county/city or by the
government.
Union. Four different unions and one category for “other or none” were
used. The unions were blue collar, white collar, academic union and the
employers organization.
Work-related psychosocial factors (the demand-control-support model)
The demand-control-support model was assessed with 19 questions in the
form of statements drawn from a modified, and for the purpose of this study,
expanded version of Karasek’s and Theorell’s instrument (Ahlberg, 1999) that
in the Swedish version contains nine questions with four alternatives. In this
study the original index was supplemented and now contains five questions
regarding work demands (work hard, too heavy burden, enough time for work
tasks – reversed score, conflicting demands, have to keep lots of information in
my mind), three questions regarding skill discretion, (my work demands a high
level of skill, I learn new things, my work demands creativity, repetitive work –
reversed score) and four questions regarding decision latitude the ability to
influence work. Each question could be answered with one of five alternatives;
“1=never”. “2=rarely”, “3=sometimes”, “4=most of the time”, and “5=always”,
where the four latter ones specified what exactly was meant by the alternative.
For each dimension a mean index was created. Groups were created for low
(mean value <2.5), medium (2.5-4.0) and high (>4.0) work demands, skill
discretion and decision latitude.
Social support was assessed with seven statements about the relation to
colleagues and supervisors (calm and pleasant atmosphere, cohesion, my
colleagues help me, It’s OK to have a bad day, I get along well with my
supervisor, I get along with my subordinates, I get along well with my
colleagues). The response scale ranged from 1-4 (disagree totally, disagree
partly, agree partly, agree totally). An index was calculated as the mean of the
10
seven questions and groups were formed with high (4.0), medium (2.6-3.9) and
low (0.0-2.5) social support.
Work hour factors
Work hours factors were assessed with 10 different variables.
Employment describes if the respondent was employed fulltime (35h/week
or more), part-time employed or self-employed. Employment status describes
whether the respondent was permanently or temporarily employed. Work hourscategorize individuals into three groups describing if they had daytime work
hours (6.45-17.45), irregular work hours without night work, and irregular work
hours with night work or permanent night work. Work shift length variationdescribes if the shifts are mostly of the same length or if they varied in length.
Influence over work hours was assessed with one question describing how often
(five categories) the respondent could decide over his/her own work hours.
Three groups were formed with a high (always), medium (often or sometimes),
and low (seldom or never) degree of influence.
Overtime was assessed with five different variables describing various
aspects of overtime. The first describes only if the respondent used to work
overtime or not. Three categories were created for the amount of paid overtimeand unpaid overtime (less than 5h, 5-15h and over15h per month). In addition,
two variables were calculated from those who worked less than 10h of totalovertime (paid and unpaid overtime) per month, or over 10h in combination with
doing it out of interest (worked overtime because work itself is enjoyable) or
being forced (have to work overtime because there are tasks to finish or because
they have been ordered to do so).
Analysis
The main interest was to relate parameters of work hour flexibility to important
endpoints of health. For this purpose was used logistic regression analysis. In the
first step a series of crude regression analyses were carried out using background
variables, psychosocial work indicators and work hour variables. The dependent
variables were the intention to remain in the work force until normal retirement;
the self-estimated inability to remain in the working force due to high mental
strain, short-term sick leave and long-term sick leave.
In a second step the analysis was controlled for background factors (model 2)
and in the third and final step (model 3), also controlled for work-related
psychosocial factors.
11
Results
The results from the logistic regression analyses are presented below under
subheadings for each of the dependent variables.
Quit working before the normal age of retirement
Table 1 below shows the results of the ´logistic regression analyses for Quitworking before the normal age of retirement. The crude analyses indicate that
not being part of one of the major unions yielded an over- risk, as did having a
low degree of social support.
Among the work hour variables, having part-time work (compared to full time
work or being self employed), having varied length of the work shifts, low
influence over work hours was associated with an over risk of quitting work
before normal age of retirement.
There was also an over risk for unpaid overtime however, an under risk for paid
overtime. Total overtime without interest or with force yielded an over risk.
When controlling for background factors and psychosocial factors, all predictors
were still significant although two more groups became significant. Having part-
time work hours (compared to full time work hours with work shifts of 7.5-9h)
and short work shifts (<7.5h) were associated with an over risk of quitting work
before normal retirement.
12
Table1
Quit work before normal retirement
Backgrond factors n OR n OR n ORGender Female 1026
male 1003 1,08 0,88 - 1,33Age 1 year 2029 1,01 1,00 - 1,02Marital status Married 1204
Not married 802 0,82 0,66 - 1,01Children < 16 No children 799
Have children 1184 0,93 0,75 - 1,15Education Elementary school 634
High school 795 1,10 0,85 - 1,41University 575 1,25 0,96 - 1,62
Sectors Manufactoring 604Health care 410 0,87 0,64 - 1,18
Service 419 0,98 0,73 - 1,32Administration 203 1,03 0,71 - 1,49
Research/developm 339 1,24 0,91 - 1,67Employer Private/self 1237
County/city 594 1,04 0,83 - 1,30Government 173 0,78 0,52 - 1,16
Union Blue Collar 602White Collar 371 1,25 0,92 - 1,70
Academic 157 1,11 0,73 - 1,69Employer 49 0,96 0,47 - 1,97
Other/none 638 1,31 1,00 - 1,70 *Demand/control + support model n OR n OR n ORWork demands Low 405 329
Medium 1299 1,02 0,78 - 1,33 1099 0,99 0,73 - 1,34High 316 1,27 0,91 - 1,79 258 1,26 0,86 - 1,85
Skill discresion High 495 418Medium 1395 0,89 0,70 - 1,13 1159 0,93 0,71 - 1,22
Low 92 0,93 0,55 - 1,57 80 1,31 0,75 - 2,29Decision latitude High 386 306
Medium 1223 0,98 0,75 - 1,28 1021 0,99 0,73 - 1,34Low 407 0,80 0,57 - 1,11 355 0,85 0,58 - 1,24
Social support High 210 173Medium 1591 1,35 0,93 - 1,94 1325 1,23 0,82 - 1,83
Low 150 2,06 1,27 - 3,36 ** 137 1,90 1,13 - 3,22 *
Work hours factors n OR n OR n OREmployment Full time 1304 1110 1079
Part time 419 1,31 1,02 - 1,7 * 337 1,59 1,16 - 2,17 ** 322 1,55 1,12 - 2,14 **Self employed 306 1,04 0,78 - 1,40 244 1,08 0,74 - 1,59 200 1,11 0,73 - 1,68
Employment status Permanent 1681 1410 1363Temporary 238 0,79 0,56 - 1,10 219 0,79 0,55 - 1,14 186 0,81 0,55 - 1,21
Work hours Daytime 1259 1065 1025Irregular no night 544 1,00 0,79 - 1,27 451 1,07 0,81 - 1,40 411 1,10 0,83 - 1,47
Irregular night 158 0,94 0,63 - 1,39 132 0,99 0,61 - 1,59 130 1,07 0,66 - 1,74Workshift length Mostly the same 1092 929 899variation Varied length 892 1,26 1,02 - 1,55 * 731 1,25 0,99 - 1,58 676 1,21 0,95 - 1,55Workshift length 7.5h-9h 872 747 727
part-time 674 1,26 0,99 - 1,59 541 1,45 1,07 - 1,95 * 485 1,48 1,08 - 2,02 *<7.5h 94 1,55 0,97 - 2,49 77 1,89 1,13 - 3,16 * 77 1,87 1,11 - 3,16 *
>9h 87 1,32 0,80 - 2,18 77 1,28 0,73 - 2,22 75 1,33 0,76 - 2,34Influence over always 226 189 160Work hours often/sometimes 1043 0,73 0,53 - 1,00 * 872 0,72 0,50 - 1,03 839 0,67 0,45 - 1,00 *
seldom/never 726 0,67 0,48 - 0,94 * 607 0,66 0,45 - 0,99 * 582 0,62 0,39 - 0,97 *
Overtime no 499 417 380yes 1458 1,12 0,88 - 1,42 1226 1,08 0,82 - 1,42 1189 1,10 0,82 - 1,47
Paid overtime <5h 1307 1087 1024per month 5-15h 446 0,76 0,59 - 0,99 * 376 0,77 0,57 - 1,02 368 0,75 0,56 - 1,01
>15h 204 0,62 0,43 - 0,91 * 180 0,68 0,45 - 1,02 177 0,65 0,43 - 0,99 *
Unpaid overtime <5h 1508 1265 1204per month 5-15h 209 1,58 1,14 - 2,18 ** 176 1,36 0,94 - 1,96 167 1,41 0,96 - 2,05
>15h 240 2,02 1,50 - 2,70 *** 202 1,81 1,29 - 2,55 *** 198 1,77 1,24 - 2,52 **
Total overtime <10h 1467 1217 1151with/without >10h no pleasure 362 1,39 1,07 - 1,80 * 308 1,41 1,05 - 1,88 * 302 1,37 1,01 - 1,85 *pleasure >10h with pleasure 128 1,32 0,88 - 1,98 118 1,12 0,71 - 1,77 116 1,04 0,65 - 1,66Total overtime <10h 1467 1217 1151with/without >10h no force 226 1,24 0,90 - 1,71 202 1,09 0,76 - 1,57 198 1,01 0,69 - 1,46force >10h with force 264 1,50 1,12 - 2,00 ** 224 1,56 1,13 - 2,15 ** 220 1,54 1,10 - 2,16 *
95% CI
Model 2
Control for background factors
95% CI
95% CI
Model 1
"Crude" logistic regression
95% CI
95% CI
95% CI
95% CI
Model 3Model 2 +
demand/control/sup.
95% CI
95% CI
13
Quit working due to mental strain
Most background factors, psychosocial factors and work hour factors studied
were associated with the feeling that work was too mentally straining to remain
in the work force until normal age of retirement (Table 2).
Being male was associated with an under risk. Having a university degree,
working with health care or research/development, not working in the private
sector and being part of a white collar or academic union was associated with an
over risk.
Being self-employed and working 5-15h of paid overtime was associated with
an under-risk for quitting work.
Over-risks were observed for irregular work hours with or without night work,
having varied shift length, working long (>9h) shifts, having low influence over
work hours, working more than 5h of unpaid over time, working more than 10h
over time without interest or with force.
When controlling for background factors and psychosocial factors only working
more than 15h of unpaid overtime, having varied work shift length and having
irregular work hours that includes night work were still significant, with an over
risk of quitting work due to mental strain.
14
Table 2
Backgrond factors n OR n OR n ORGender Female 1019
male 983 0,61 0,49 - 0,75 ***
Age 1 year 2002 1,01 1,00 1,02Marital status Married 1186
Not married 793 0,99 0,79 - 1,23Children < 16 No children 799
Have children 1161 1,19 0,95 - 1,49Education Elementary school 612
High school 793 1,14 0,86 - 1,50University 573 2,11 1,59 - 2,79 ***
Sectors Manufactoring 593Health care 410 2,98 2,18 - 4,05 ***
Service 413 1,19 0,84 - 1,69Administration 202 0,70 0,42 - 1,17
Research/developm 339 2,61 1,88 - 3,62 ***
Employer Private/self 1215County/city 591 2,81 2,23 - 3,55 ***Government 173 1,50 1,01 - 2,24 *
Union Blue Collar 590White Collar 371 1,64 1,20 - 2,24 **
Academic 157 1,89 1,27 - 2,83 **Employer 48 1,05 0,50 - 2,24
Other/none 629 1,03 0,77 - 1,38Demand/control + support model n OR n OR n ORWork demands Low 396 326
Medium 1296 3,16 2,10 - 4,75 *** 1098 3,15 1,96 - 5,07 ***High 301 9,99 6,39 - 15,6 *** 250 9,41 5,58 - 15,8 ***
Skill discresion High 491 417Medium 1385 0,72 0,56 - 0,91 ** 1155 0,83 0,62 - 1,10
Low 94 0,57 0,32 - 1,02 82 0,93 0,50 - 1,76Decision latitude High 369 298
Medium 1217 1,52 1,11 - 2,10 ** 1019 1,35 0,93 - 1,97Low 406 2,06 1,43 - 2,96 *** 355 2,04 1,32 - 3,14 ***
Social support High 200 165Medium 1583 5,62 2,85 - 11,1 *** 1322 4,32 2,15 - 8,66 ***
Low 148 16,1 7,67 - 33,9 *** 137 14,7 6,78 - 31,8 ***
Work hours factors n OR n OR n OREmployment Full time 1291 1103 1078
Part time 417 1,24 0,95 - 1,60 337 0,98 0,70 - 1,36 321 1,01 0,70 - 1,44Self employed 294 0,61 0,43 - 0,87 ** 239 0,77 0,48 - 1,21 198 0,79 0,47 - 1,33
Employment status Permanent 1667 1404 1362Temporary 234 0,70 0,48 - 1,01 215 0,79 0,53 - 1,19 185 0,87 0,55 - 1,39
Work hours Daytime 1243 1059 1023Irregular no night 539 1,39 1,09 - 1,77 ** 448 1,38 1,03 - 1,85 * 410 1,24 0,90 - 1,70
Irregular night 156 1,90 1,31 - 2,75 *** 131 1,72 1,09 - 2,70 * 129 1,63 1,00 - 2,65 *
Workshift length Mostly the same 1081 924 898variation Varied length 879 1,72 1,38 - 2,15 *** 726 1,71 1,32 - 2,21 *** 674 1,58 1,20 - 2,07 ***
Workshift length 7.5h-9h 865 742 727part-time 660 1,06 0,82 - 1,36 536 0,94 0,67 - 1,30 482 0,99 0,69 - 1,41
<7.5h 94 1,13 0,67 - 1,90 77 1,05 0,59 - 1,87 77 1,02 0,55 - 1,87>9h 84 1,67 1,01 - 2,76 * 75 1,35 0,76 - 2,40 74 1,31 0,71 - 2,42
Influence over always 217 186 157Work hours often/sometimes 1036 1,44 0,95 - 2,16 868 1,23 0,78 - 1,94 840 0,91 0,54 - 1,55
seldom/never 718 1,99 1,31 - 3,02 *** 603 1,47 0,90 - 2,38 581 0,98 0,55 - 1,74Overtime no 489 412 379
yes 1449 1,27 0,98 - 1,65 1222 1,38 1,02 - 1,87 * 1188 1,06 0,76 - 1,48Paid overtime <5h 1292 1080 1022per month 5-15h 443 0,72 0,55 - 0,96 * 375 0,86 0,62 - 1,17 368 0,72 0,52 - 1,01
>15h 203 1,01 0,71 - 1,44 179 1,16 0,77 - 1,75 177 0,82 0,53 - 1,27Unpaid overtime <5h 1491 1257 1201per month 5-15h 209 1,93 1,39 - 2,66 *** 176 1,62 1,10 - 2,37 * 168 1,49 0,98 - 2,25
>15h 238 2,06 1,52 - 2,79 *** 201 1,98 1,36 - 2,89 *** 198 1,54 1,03 - 2,32 *
Total overtime <10h 1451 1210 1149with/without >10h no pleasure 359 1,82 1,41 - 2,37 *** 306 1,90 1,40 - 2,59 *** 302 1,32 0,95 - 1,83pleasure >10h with pleasure 128 0,74 0,45 - 1,22 118 0,72 0,40 - 1,27 116 0,61 0,33 - 1,12Total overtime <10h 1451 1210 1149with/without >10h no force 226 1,00 0,70 - 1,42 202 1,03 0,69 - 1,56 198 0,84 0,55 - 1,31force >10h with force 261 2,01 1,50 - 2,69 *** 222 2,05 1,45 - 2,90 *** 220 1,37 0,95 - 1,99
Quit work before normal retirement due to mental strain
95% CI
95% CI
95% CI
Model 1
"Crude" logistic regression
95% CI95% CI
Model 2
Control for background factors
95% CI
95% CI
Model 3Model 2 +
demand/control/sup.
95% CI
95% CI
15
Short-term sick leave
Table 3 shows the results for short-term sick leave (over 24 days/year).
Being a male, being young and being an employer were associated with
an under risk of short-term sick leave in the crude analyses.
Being single, having a high school education, working for the
county/city, having low social support and low decision latitude were
associated with an increased risk
Significant work hour factors associated with an under risk were: being
self employed, having a temporary employment status, varied length of
work shifts, working part-time (compared to full-time with 7.5-9h shifts)
and working over 10h overtime in combination with over time out of
interest. Over-risk for short-term sick leave was observed only for a low
influence over work hours.
When controlling for background factors, all work hour predictors lost
their significance. However, part-time work (compared to full-time
work) became significant with an over risk. This over-risk, was still
significant after controlling for psychosocial factors.
16
Table 3
Short-term sick leave (>24days)
Backgrond factors n OR n OR n ORGender Female 1006
male 1012 0,68 0,57 - 0,81 ***
Age 1 year 2018 0,97 0,97 - 0,98 ***
Marital status Married 1190Not married 805 1,36 1,13 - 1,63 ***
Children < 16 No children 795Have children 1179 1,29 0,93 - 1,78
Education Elementary school 635High school 793 1,32 1,07 - 1,62 *
University 566 1,23 0,98 - 1,55Sectors Manufactoring 608
Health care 402 1,11 0,86 - 1,43Service 408 1,03 0,80 - 1,32
Administration 200 1,02 0,74 - 1,41Research/developm 336 1,15 0,88 - 1,50
Employer Private/self 1231County/city 586 1,25 1,02 - 1,52 *Government 169 1,15 0,83 - 1,59
Union Blue Collar 596White Collar 360 0,92 0,71 - 1,20
Academic 154 0,77 0,54 - 1,10Employer 51 0,31 0,17 - 0,57 ***
Other/none 632 0,79 0,63 - 0,99 *Demand/control + support model n OR n OR n ORWork demands Low 403 328
Medium 1280 1,17 0,94 - 1,47 1086 1,27 0,78 - 2,05High 313 1,26 0,94 - 1,70 254 1,85 1,03 - 3,32 *
Skill discresion High 485 409Medium 1379 1,16 0,94 - 1,42 1150 0,97 0,63 - 1,47
Low 93 1,45 0,92 - 2,28 79 1,26 0,58 - 2,72Decision latitude High 387 308
Medium 1202 1,66 1,32 - 2,10 *** 1007 1,79 1,00 - 3,20 *Low 408 2,52 1,89 - 3,36 *** 352 2,12 1,12 - 4,00 *
Social support High 213 177Medium 1574 1,36 1,02 - 1,81 * 1306 1,10 0,59 - 2,03
Low 147 2,10 1,36 - 3,24 *** 137 1,79 0,83 - 3,84Work hours factors n OR n OR n OREmployment Full time 1297 1098 1068
Part time 410 1,09 0,87 - 1,4 332 1,59 1,04 - 2,43 * 317 1,67 1,07 - 2,61 *Self employed 311 0,46 0,36 - 0,59 *** 244 1,11 0,57 - 2,17 199 1,11 0,53 - 2,32
Employment status Permanent 1659 1391 1346Temporary 239 0,61 0,46 - 0,80 *** 219 0,59 0,30 - 1,16 185 0,66 0,32 - 1,36
Work hours Daytime 1242 1047 1007Irregular no night 543 0,87 0,71 - 1,06 454 0,88 0,57 - 1,34 411 0,79 0,50 - 1,24
Irregular night 157 0,93 0,67 - 1,30 131 1,07 0,56 - 2,02 130 1,04 0,54 - 2,00Workshift length Mostly the same 1088 919 888variation Varied length 877 0,75 0,63 - 0,90 ** 726 0,92 0,64 - 1,32 670 0,87 0,59 - 1,29Workshift length 7.5h-9h 865 736 718
part-time 668 0,69 0,56 - 0,84 *** 534 1,40 0,90 - 2,15 477 1,53 0,97 - 2,41<7.5h 97 0,74 0,49 - 1,13 79 0,72 0,30 - 1,77 79 0,75 0,30 - 1,87
>9h 89 0,72 0,46 - 1,11 79 1,06 0,43 - 2,62 76 1,15 0,46 - 2,88Influence over always 227 192 160Work hours often/sometimes 1035 1,59 1,19 - 2,12 ** 867 0,74 0,39 - 1,39 834 0,66 0,32 - 1,35
seldom/never 711 1,70 1,26 - 2,30 *** 593 1,40 0,74 - 2,67 571 1,15 0,55 - 2,44Overtime no 494 409 372
yes 1443 1,04 0,84 - 1,27 1219 0,82 0,56 - 1,21 1181 0,72 0,48 - 1,09Paid overtime <5h 1298 1079 1014per month 5-15h 440 1,12 0,90 - 1,39 374 0,96 0,62 - 1,49 367 0,88 0,56 - 1,38
>15h 199 1,22 0,90 - 1,65 175 0,80 0,41 - 1,55 172 0,72 0,37 - 1,42Unpaid overtime <5h 1496 1254 1191per month 5-15h 208 0,99 0,74 - 1,33 177 0,79 0,40 - 1,53 168 0,75 0,37 - 1,52
>15h 233 0,78 0,59 - 1,03 197 1,55 0,88 - 2,73 194 1,56 0,86 - 2,81Total overtime <10h 1458 1210 1142with/without >10h no pleasure 353 1,15 0,91 - 1,45 302 1,22 0,76 - 1,94 297 1,11 0,69 - 1,80pleasure >10h with pleasure 126 0,67 0,46 - 0,96 * 116 0,37 0,11 - 1,20 114 0,39 0,12 - 1,27Total overtime <10h 1458 1210 1142with/without >10h no force 228 0,91 0,69 - 1,21 203 0,91 0,49 - 1,68 199 0,90 0,48 - 1,69force >10h with force 251 1,07 0,81 - 1,40 215 1,06 0,60 - 1,87 212 0,96 0,53 - 1,72
95% CI
Model 2
Control for background factors
95% CI
95% CI
Model 1
"Crude" logistic regression
95% CI
95% CI
95% CI
95% CI
Model 3Model 2 +
demand/control/sup.
95% CI
95% CI
17
Long-term sick leave
The results from analyses of long-term sick leave (currently on sick leave for
more than 3 months) are presented in table 4. The crude regression analyses
show that being a male was associated with an under risk. Over-risks were
observed for higher age, having children, working in the health care sector or
service sector, being employed in the county/city, having high work demands,
low decision latitude and low social support.
Varied work shift length, working overtime and working overtime more than
10h without being forced was associated with an under risk of long-term sick
leave. An over risk was observed for having a low influence over work hours.
When controlling for background factors and psychosocial factors, the under
risk for varied length of work shifts and working overtime without force were
still significant. In addition, having a temporary employment status was
associated with an increased risk of long-term sick leave.
18
Table 4
Long-term sick leave (>3months)
Backgrond factors n OR n OR n ORGender Female 1125
male 1092 0,31 0,20 - 0,50 ***
Age 1 year 2217 1,04 1,03 - 1,06 ***
Marital status Married 1316Not married 871 0,93 0,61 - 1,41
Children < 16 No children 837Have children 1326 2,12 1,32 - 3,41 **
Education Elementary school 738High school 846 0,77 0,47 - 1,25
University 601 0,92 0,55 - 1,53Sectors Manufactoring 660
Health care 442 2,62 1,39 - 4,92 **Service 459 2,12 1,11 - 4,06 *
Administration 216 1,75 0,76 - 4,02Research/development 359 2,00 1,00 - 4,01
Employer Private/self 1346County/city 640 1,93 1,26 - 2,98 *Government 187 1,26 0,59 - 2,72
Union Blue Collar 663White Collar 384 0,69 0,37 - 1,27
Academic 163 0,76 0,33 - 1,74Employer 54 0,01 0,00 - ###
Other/none 693 0,66 0,40 - 1,10Demand/control + support model n OR n OR n ORWork demands Low 443 347
Medium 1368 1,31 0,69 - 2,48 1130 4,45 0,57 - 34,8High 365 3,57 1,82 - 7,02 *** 274 6,10 0,69 - 53,6
Skill discresion High 527 431Medium 1486 0,87 0,53 - 1,41 1198 1,03 0,39 - 2,69
Low 104 1,06 0,39 - 2,84 83 0,00 0,00 - ###Decision latitude High 421 328
Medium 1305 2,23 1,05 - 4,71 * 1055 5,18 0,65 - 41,3Low 448 3,70 1,68 - 8,16 *** 366 5,86 0,65 - 53,0
Social support High 231 187Medium 1689 1,43 0,61 - 3,34 1365 2,52 0,32 - 19,6
Low 174 5,71 2,27 - 14,4 *** 141 4,91 0,52 - 46,6Work hours factors n OR n OR n OREmployment Full time
Part timeSelf employed
Employment Permanent 1792 1455 1432status Temporary 279 1,37 0,79 - 2,39 240 0,57 0,12 - 2,72 208 2,63 1,27 - 5,45 **
Work hours Daytime 1348 1101 1076Irregular no night 600 1,40 0,89 - 2,21 477 1,51 0,60 - 3,83 441 1,23 0,67 - 2,25
Irregular night 168 1,25 0,58 - 2,67 136 1,31 0,25 - 6,93 137 0,76 0,28 - 2,07Workshift length Mostly the same 1186 964 956variation Varied length 959 0,62 0,40 - 0,96 * 764 0,49 0,19 - 1,22 710 0,53 0,31 - 0,93 *
Workshift length 7.5h-9hpart-time
<7.5h>9h
Influence over always 250 205 169Work hours often/sometimes 1110 1,36 0,57 - 3,27 901 0,53 0,13 - 2,23 882 1,08 0,30 - 3,84
seldom/never 796 2,84 1,21 - 6,70 * 631 0,89 0,22 - 3,68 622 1,23 0,34 - 4,46Overtime no 554 440 408
yes 1552 0,55 0,36 - 0,86 ** 1268 0,70 0,27 - 1,80 1253 0,59 0,34 - 1,04Paid overtime <5h 1414 1135 1083per month 5-15h 470 0,76 0,43 - 1,32 386 0,72 0,20 - 2,56 388 0,80 0,41 - 1,54
>15h 222 0,91 0,44 - 1,85 187 0,57 0,07 - 4,49 190 0,78 0,33 - 1,89Unpaid overtime <5h 1639 1321 1283per month 5-15h 219 0,49 0,20 - 1,22 181 0,00 0,00 - ### 175 0,37 0,12 - 1,13
>15h 248 0,70 0,33 - 1,46 206 1,85 0,59 - 5,78 203 0,71 0,28 - 1,77Total overtime <10h 1590 1271 1225with/without >10h no pleasure 385 0,87 0,49 - 1,53 318 2,02 0,73 - 5,61 318 0,79 0,40 - 1,57pleasure >10h with pleasure 131 0,33 0,08 - 1,37 119 0,00 0,00 - ### 118 0,25 0,03 - 1,93Total overtime <10h 1590 1271 1225with/without >10h no force 234 0,19 0,05 - 0,76 * 207 0,00 0,00 - ### 204 0,11 0,02 - 0,86 *force >10h with force 282 1,20 0,68 - 2,13 230 2,99 1,03 - 8,68 * 232 1,15 0,56 - 2,34
95% CI
95% CI
95% CI
Model 1
"Crude" logistic regression
95% CI95% CI
Model 2
Control for background factors
95% CI
95% CI
Model 3Model 2 +
demand/control/sup.
95% CI
95% CI
19
DiscussionThe main impression of the results is that work hours seem to be important for
all the variables analyzed in this study, but to a varied degree. However, before
the results are further discussed a short remark on the statistical analyses has to
be made.
When comparing the three models of regression analyses for the dependent
variables it is important to keep in mind that there are two main explanations for
the loss of significance when control variables are entered into the analyses.
First, control variables may share some of the variance with the independent
variables and therefore the independent variables loose significance. If this
would be the main explanation for loss of significance then the odds ratios for
the independent variables should also be substantially affected. Second, the
more variables that are entered in to the analyses the lower is the statistical
power. This is partly due to loss of subjects but also because of loss of degrees
of freedom due to more independent variables in the analyses. This means that a
correct interpretation of the results has to take into account the changes in odds
ratios across the different models as well as the number of independent variables
and loss of subjects.
Among the predictors, part-time work was associated with an increased risk of
short-term sick leave and a risk of quitting work before normal age of
retirement. In addition, short work shift (<7.5h) when working full-time work
hours increased the risk for quitting work before normal retirement. Part-time
work would perhaps at first thought be associated with a better prognosis for
survival in the work force. However, it is probably the case that shorter work
days and/or work hours to some extent is a coping strategy for less healthier
individuals and/or individuals with a high total workload (in and outside work).
One of the reasons for having part-time work is often a health problem or social
obligations at home. Thus, this group would be more likely to quit work early in
life if they have the opportunity to, and also to have more sickness absence
while still in the work force.
Temporary employment status was associated with an over risk for long-term
sick leave but, an under risk for short-term sick leave. In the latter case,
significance was lost after controlling for background factors, although, odds
ratios were hardly affected. A plausible explanation is that the loss of
significance had more to do with loss of statistical power than the variance
shared with the control variables. Interestingly, there is a substantial increase in
the risk of having long-term sick leave (from 0.57 to 2.63) after introduction of
psychosocial factors for the group that has a temporary employment status.
Apparently, there is a synergistic interaction with other psychosocial factors at
work. Having a permanent position increases the risk for long-term sick leave
the most when other psychosocial factors are taken into account.
20
Irregular work hours with night work and work hours with varied shift length
were associated with quitting work due to mental strain. This indicates that
irregular working hours contributed to mental strain. Varied shift length was
associated with an under risk of long-term sick leave. The result contrasts to the
increased risk of irregular work-hours discussed above. However, other aspects
of irregular work-hours did not show an under risk.
Low influence over work-hours was associated with a higher risk for long-term
& short-term sick leave and to quit work due to mental strain in the crude
analyses. However, the predictors lost their significance when controlled for
psychosocial factors. In these cases there are also a decrease in odds ratios,
which indicates that the control variables account for part of the variance. This
was not very surprising because influence over work hours is closely related to
work demands and decision latitude. High demands decrease influence and high
decision latitude is also often related to work hours as well as to the work
situation.
Low influence over work hours also indicated an under risk for quitting work
before normal age of retirement. The reason for this under risk is not clear. A
possible explanation is that quitting work before normal retirement also is
influenced by economic factors. It is plausible that individuals who have the
economic possibility to quit work early in life also have more influence over
work hours.
Overtime work seems to be mainly associated with an under risk for sick leave
(both long-term and short-term), especially in combination with interest or in the
absence of force. However, when controlling for background and psychosocial
factors only the combination overtime without force was significant as a
predictor and only for long-term sick leave.
When examining the odds ratios for overtime work in the analyses above, loss of
statistical power seems to be the most plausible explanation for the predictors to
lose their significance. Odds ratios are hardly affected by the control variables
but there is a loss of subjects in the analyses when more control variables are
entered.
A similar pattern was also shown for variables describing if one would work
until normal retirement (both due to mentally straining work and for any
reason). However, in these cases, overtime was associated with an over risk in
combination with the absence of interest or in combination with force. However,
when controlling for background factors and psychosocial factors significance is
lost for the variable describing if one would quit before normal retirement due to
mental strain. The main reason for loss of significance seem to be that the
control variables accounted for the variance, because odds ratios were
substantially decreased when psychosocial factors were entered in to the
analyses. Apparently, overtime in combination with force or pleasure are to
some extent explained by the demand-control-support model.
In the prediction of who will quit work before normal retirement (sp. for any
reason) the combination of overtime without interest or with force was,
21
however, still significant after controlling for psychosocial factors. Apparently,
overtime in combination with force or the absence of pleasure has a more robust
effect on this outcome.
There also seems to be differential effects of the amount of overtime and
whether the overtime is paid for or not. Unpaid overtime seems to increase the
risk for quitting work before normal retirement but paid overtime seem to
decrease the risk. When controlling for background and psychosocial factors the
under risk for paid overtime loose significance in predicting who’s quitting work
due to mental strain. But again, the explanations seem to have more to do with
loss of statistical power than the control variables per se.
Overtime seem to be a complex variable and different kinds of overtime seem to
have differential effects. Paid overtime with pleasure and the absence of being
forced to work overtime seem to be mainly beneficial. Unpaid overtime,
especially in combination with force or the absence of pleasure seem have more
negative effects.
22
ReferencesCosta, G. Guidelines for the medical surveillance of shift workers. Scand J
Work Environ Health 1998,24:151-155.
Costa, G. Åkerstedt, T. Nachreiner, F., et al. Flexible work hours, health and
well being in the Europen Union: preliminary data from a SALTSA projekt. JHum Ergol in press.
Golembiewski, R. T. Hillies, R. and Kagno, M. S. A longitudinal study of
flex-time effects: Some consequences of an OD structural intervention. J ApplBehav Sci 1974,10:503-532.
Knauth, P. Innovative worktime arrangements. Scand J Work Environ Health1998,24:13-17.
Knauth, P. Kiesswetter, E. and Schwarzenau, P. Erfahrungen mit einer
flexiblen Arbeitsregelung bei Dreischichtarbeitern. Z. Arb.wiss 1984,38:96-99.
Knutsson, A. and Bøggild, H. Shiftwork and cardiovascular disease: review
of disease mechanisms. Rev Environ Health 2000,15:359-72.
Kogi, K. Increasing flexibility in shiftwork arrangements. Work and Stress1995,9:211-218.
Kogi, K. Implications of flexible work systems for work studies. J Hum Ergol1997,26:89-97.
McEwan Young, W. Shift work and flexible schedules: are they compatible?
Int Labour Rev 1980,119:1-17.
Narayanan, V. K. and Nath, R. Hierarchical Level and the Impact of Flextime.
Industrial Relations 1982,21:216-230.
Tepas, D. I. Flexitime, compressed workweeks and other alternative work
schedules. In: S. Folkard, T.H. Monk, (Eds). Hours of Work. John Wiley &
Sons, Chichester, 1985:147-164.
Torsvall, L. and Åkerstedt, T. Effects of flexible working hours, parttime
working and fourday workweek. 58b, Reports from the Laboratory for clinical
stress research, 1977.
Wade, M. eds. Flexible working hours in practice. Wiley, New York, 1974.
Åkerstedt, T. Shift work and disturbed sleep/wakefulness. Sleep MedicineReviews 1998,2:117-128.
StressforskningsinstitutetStockholms universitet 106 91 Stockholm www.stressforskning.su.se
Stressforskningrapporter 2003-2013325 Tucker P, Bejerot E, Kecklund G, Aronsson G & Åkerstedt T (2013) Doctors’ work hours in Sweden: Their impact on sleep, health, work-family balance, patient care and thoughts about work.324 Åkerstedt T, Ingre M, Kecklund G (2012) Vad kännetecknar bra och dåliga skiftscheman?323 Lowden A, Åkerstedt T (2012) Ljus i kontrollrummet vid Forsmark 3 anpassat till skiftschema för optimering av synergonomi, vakenhet och återhämtning.322 Kecklund G, Ingre M, Åkerstedt T (2010) Arbetstider, hälsa och säkerhet – en uppdatering av aktuell forskning.321 Kinsten A, Magnusson Hanson L, Hyde M, Oxenstierna G, Westerlund H, Theorell T (2007) SLOSH – Swedish Longitudinal Occupational Survey of Health: a nationally representative psychosocial survey of the Swedish working population.320 Oxenstierna G, Widmark M, Finnholm K, Elofsson S (2008) Psykosociala faktorer i dagens arbetsliv och hur man mäter och beskriver dem.319 Kecklund G, Eriksen CA, Åkerstedt T (2006) Hälsa, arbetstider och säkerhet inom polisen. 318 Thulin Skantze E (2006) Organisationsstrukturens betydelse för de anställdas hälsa. En explorativ studie baserad på fokusgruppsintervjuer bland chefer.317 Söndergaard HP (2006) Hälsoeffekter av rån och övriga traumatiska händelser bland handelsanställda. Delstudie I och II.316 Holmén-Isaksson M (2005) Rehabilitering. En förhandling mellan individer och organisationer.315 Widmark M (2005) Det nya arbetslivet. En explorativ studie som jämför två dominerande psykosociala arbetsmiljömodeller med aktuell arbetsmiljöproblematik iorganisations- förhållanden.314 Hasselhorn HM, Theorell T, Hammar N, Alfredsson L, Westerholm P and the WOLF Study Group (2004) Occupational health care team ratings and self reports of demands and decision latidude – results from the Swedish WOLF Study.313 Bernin P, Theorell T (2004) Mönster för framgångsrikt ledarskap i vården.312 Söderström M, Jeding K, Ekstedt M, Kecklund G, Åkerstedt T (2003) Arbetsmiljö, stress och utbrändhet inom ett företag i IT-branschen.311 Hansen Egon (2003) Stress, Stream of Affect, Emotions, and Background Variables: Exploratory Experiment with Poetry Reading II.310 Pernler H, Gillberg M (2003) Sömnvanor och sömnproblem hos barn i förskoleåldern.